yambs

README

yambs (Yet Another MusicBrainz Seeder)

yambs is a command-line program for seeding edits to the MusicBrainz music database.

It can simplify adding multiple standalone recordings: given a CSV or TSV file describing recordings, yambs can open the Add Standalone Recording page for each with various fields pre-filled.

yambs can also read key=value lines from text files to seed the Add Release page, and it can use Bandcamp album pages and local MP3 files to seed edits too.

There's a web frontend at yambs.erat.org.

Installation

To compile and install the yambs executable, install Go and run the following command:

go install ./cmd/yambs

Usage

Usage yambs: [flag]... <FILE/URL>
Seeds MusicBrainz edits.

  -action value
    	Action to perform with seed URLs (open, print, serve, write) (default open)
  -addr string
    	Address to listen on for -action=serve (default "localhost:8999")
  -fields string
    	Comma-separated fields for CSV/TSV columns (e.g. "artist,name,length")
  -format value
    	Format for text input (csv, keyval, tsv) (default tsv)
  -list-fields
    	Print available fields for -type and exit
  -set value
    	Set a field for all entities (e.g. "edit_note=from https://www.example.org")
  -type value
    	Entity type for text or MP3 input (recording, release)
  -verbose
    	Enable verbose logging
  -version
    	Print the version and exit

yambs reads the supplied file or URL (or stdin if no positional argument is supplied) and performs the action specified by the -action flag:

• open: Open edits in a browser using a temporary file.
• print: Write edit links to stdout (only possible for recordings).
• serve: Open edits in a browser using a short-lived webserver launched at -addr (useful if you're running yambs in a container).
• write: Write a webpage containing the edits to stdout.

If you supply a URL, yambs will fetch and parse it.

If you supply a filename, you should also pass the -type, -format, -fields, and -set flags to tell yambs how to interpret the file.

Examples

To add multiple non-album recordings for a single artist, you can run a command like the following:

yambs \
  -type recording \
  -format tsv \
  -fields name,length,edit_note \
  -set artist=7e84f845-ac16-41fe-9ff8-df12eb32af55 \
  -set url0_url=https://www.example.org/ \
  -set url0_type=255 \
  <recordings.tsv

with a recordings.tsv file like the following (with tab characters between the fields):

Song #1	4:35	info from https://example.org/song1.html
Song #2	53234.35	info from https://example.org/song2.html

The recordings' names, lengths, and edit notes will be read from the TSV file, and the -set artist=... flag sets all recordings' artist field to the specified artist.

Likewise, the -set url0_... flags add a URL relationship to each recording. seed/enums.go enumerates the different link types that can be specified between entities; 255 corresponds to LinkType_DownloadForFree_Recording_URL.

---

To edit existing recordings, specify their MBIDs via the mbid field:

yambs \
  -type recording \
  -format csv \
  -fields mbid,name \
  <recordings.csv

recordings.csv:

c55e74ff-bd7d-40ff-a591-c6993c59bda8,Sgt. Pepper’s Lonely Hearts Club Band
...

Note that this example uses the csv format rather than tsv.

---

More-complicated artist credits can also be assigned:

yambs \
  -type recording \
  -format tsv
  -fields ... \
  -set artist0_mbid=1a054dd8-c5fa-40b6-9397-61c26b0185d4 \
  -set artist0_credited=virt \
  -set 'artist0_join= & ' \
  -set artist1_name=Rush \
  ...

(Note that repeated fields are 0-indexed.)

---

The keyval format can be used to seed a single entity across multiple lines:

yambs \
  -type release \
  -format keyval \
  <release.txt

release.txt:

title=Some Album
artist0_name=Some Artist
types=Album,Soundtrack
status=Official
packaging=Jewel Case
language=eng
script=Latn
event0_date=2021-05-15
event0_country=XW
medium0_format=CD
medium0_track0_title=First Track
medium0_track0_length=3:45.04
medium0_track1_title=Second Track
medium1_format=CD
medium1_track0_title=First Track on Second Disc
url0_url=https://www.example.org/
url0_type=75
edit_note=https://www.example.org

seed/enums.go shows that the url0_type=75 line corresponds to LinkType_DownloadForFree_Release_URL.

---

Pass the -list-fields flag to list all available fields for a given entity type:

yambs -type recording -list-fields
yambs -type release   -list-fields

Acceptable values for various fields are listed in seed/enums.go, which is automatically generated from t/sql/initial.sql in the musicbrainz-server repository.

A column in an input file can be assigned to multiple fields by supplying slash-separated field names. For example, -fields name,url0_url/edit_note,length maps the first column to field name, the second column to fields url0_url and edit_note, and the third column to field length.

A column can be skipped by passing an empty field name. For example, -fields name,,length maps the first column to field name, skips the second column, and maps the third column to length.

---

You can pass Bandcamp album URLs to seed release edits:

yambs https://austinwintory.bandcamp.com/album/journey

The page that is opened will include a link to the album's highest-resolution cover art to make it easier to add in a followup edit.

If you pass a Bandcamp track URL that isn't part of an album, an edit to add it as a single will be created:

yambs https://caribouband.bandcamp.com/track/tin

---

You can pass the path to a local MP3 file to use it to seed a (single) release or standalone recording edit:

yambs \
  -type recording \
  -set artist=7e84f845-ac16-41fe-9ff8-df12eb32af55 \
  -set edit_note='from artist-provided MP3 at https://www.example.org/song.mp3' \
  /path/to/a/song.mp3

If the MP3 file contains embedded images, they will be extracted to temporary files so they can be added as cover art.

---

There's also a yambsd executable that exposes most of the same functionality through a webpage (with some limits to avoid abuse).

Why?

There are a bunch of MusicBrainz userscripts that run in the browser with the help of an extension like Tampermonkey to seed edits. They're well-tested, so why not just use them instead of writing a new thing?

Well, at first I was adding a bunch of standalone recordings that I'd downloaded from random musicians' homepages. I couldn't find any userscripts to help with that, since the main focus seems to be seeding releases from major websites. I ended up hacking together a shell script to generate URLs that would seed my edits, but I figured it'd be nice to have something more robust and convenient to use next time.

I had also been using the bandcamp_importer.user.js userscript to import releases from Bandcamp, but I'm nervous about using extensions like Tampermonkey that require permission to modify data on all sites. I'm not so worried about malice on the part of extension or userscript developers, but I have no idea about their security practices and I'm fearful of attackers compromising their computers and uploading malicious versions of their code.

I created a separate browser profile that I could use to run Tampermonkey without exposing any of my (non-MusicBrainz) credentials, but using it was a pain, so I decided to add Bandcamp support to this codebase as well since that's where I get most of my music.

Further reading

• "Release Editor Seeding" documentation
• Release seeding example
• "Seeding Recordings" thread